Detecting content of interest in streaming media

Abstract

The present invention recognizes media content using signatures generated by network devices with limited processing power. A signature comprises a first set of frequency-amplitude pairs associated with a first piece of content. The signature is compared with stored signatures corresponding to known content. Each of the stored signatures comprises a stored set of frequency-amplitude pairs. The first piece of content is identified as being or including known content in response to the first signature matching a stored signature.

What is claimed is: 
     
         1 . An automated process executable on a client device to playback media content, the automated process comprising:
 receiving, by the client device, the media content for playback;   generating, by the client device, a first signature of the media content at a playback time, the first signature comprising a first set of frequency-amplitude pairs associated with the media content,
 wherein the client device applies a window function to generate tapered amplitudes, 
 wherein the client device applies a discrete Fourier transform (DFT) to the tapered amplitudes to generate outputs for frequency bins, and 
 wherein the client device selects dominant output magnitudes from the outputs to generate the first set of frequency-amplitude pairs; 
   transmitting, by a client device and to a server, the first signature, wherein the server is configured to detect that the media content includes a known content in response to the first signature matching a stored signature of the known content;   receiving, by the client device and from the server, replacement content in response to the server detecting that the media content includes the known content; and   playing, by the client device, the media content including the replacement content at the playback time.   
     
     
         2 . The automated process of  claim 1 , further comprising scaling, by the client device and proportionally, input amplitudes of an audio component of the media content into integers. 
     
     
         3 . The automated process of  claim 2 , wherein the integers comprise 16-bit integers or 32-bit integers. 
     
     
         4 . The automated process of  claim 2 , wherein the client device applies a window function to scaled amplitudes using a trigonometric lookup table to generate the tapered amplitudes. 
     
     
         5 . The automated process of  claim 1 , wherein the first signature matches the stored signature in response to the first set of frequency-magnitude pairs being the same as the stored set of frequency-magnitude pairs over a predetermined duration. 
     
     
         6 . The automated process of  claim 5 , wherein the predetermined duration is between three seconds and five seconds. 
     
     
         7 . The automated process of  claim 1 , wherein the first signature comprises the first set of frequency-magnitude pairs associated with a first window of the media content and a second set of frequency-magnitude pairs associated with a second window of the media content. 
     
     
         8 . The automated process of  claim 7 , wherein a duration of the first window is in a range from 50 milliseconds to 150 milliseconds. 
     
     
         9 . The automated process of  claim 8 , wherein the first window and the second window overlap. 
     
     
         10 . The automated process of  claim 7 , wherein the first set of frequency-magnitude pairs comprises five frequency-magnitude pairs selected in response to having dominant magnitudes in the first window. 
     
     
         11 . A client device configured to playback media and comprising a processor and a non-transitory, computer readable media configured to store instructions thereon that, when executed by the processor, cause the client device to perform operations, the operations comprising:
 receiving, by a client device, media content for playback;   generating, by the client device, a first signature of the media content at a playback time, the first signature comprising a first set of frequency-amplitude pairs associated with the media content,
 wherein the client device applies a window function to generate tapered amplitudes, 
 wherein the client device applies a discrete Fourier transform (DFT) to the tapered amplitudes to generate outputs for frequency bins, and 
 wherein the client device selects dominant output magnitudes from the outputs to generate the first set of frequency-amplitude pairs; 
   transmitting, by a client device and to a server, the first signature, wherein the server is configured to detect that the media content includes a known content in response to the first signature matching a stored signature of the known content;   receiving, by the client device and from the server, replacement content in response to the server detecting that the media content includes the known content; and   playing, by the client device, the media content including the replacement content at the playback time.   
     
     
         12 . The client device of  claim 11 , wherein the operations further comprise scaling, by the client device and proportionally, input amplitudes of an audio component of the media content into integers. 
     
     
         13 . The client device of  claim 12 , wherein the client device applies a window function to scaled amplitudes using a trigonometric lookup table to generate the tapered amplitudes. 
     
     
         14 . The client device of  claim 11 , wherein the first signature comprises the first set of frequency-magnitude pairs associated with a first window of the media content and a second set of frequency-magnitude pairs associated with a second window of the media content. 
     
     
         15 . The client device of  claim 14 , wherein a duration of the first window is in a range from 50 milliseconds to 150 milliseconds. 
     
     
         16 . The client device of  claim 15 , wherein the first window and the second window overlap. 
     
     
         17 . The client device of  claim 14 , wherein the first set of frequency-magnitude pairs comprises five frequency-magnitude pairs selected in response to having dominant magnitudes in the first window. 
     
     
         18 . An automated process executable on a client device to playback media content, the automated process comprising:
 generating, by the client device, a first signature of the media content at a playback time, the first signature comprising a first set of frequency-amplitude pairs associated with the media content,
 wherein the client device applies a window function and a discrete Fourier transform (DFT) to generate outputs for frequency bins, and 
 wherein the client device selects dominant output magnitudes from the outputs to generate the first set of frequency-amplitude pairs; 
   transmitting, by a client device and to a server, the first signature;   receiving, by the client device, replacement content in response to transmitting the first signature to the server; and   playing, by the client device, the media content including the replacement content at the playback time.   
     
     
         19 . The automated process of  claim 18 , further comprising scaling, by the client device and proportionally, input amplitudes of an audio component of the media content into integers. 
     
     
         20 . The automated process of  claim 18 , wherein generating the first signature further comprises applying, by the client device, a window function to scaled amplitudes using a trigonometric lookup table to generate tapered amplitudes.